A board member asks for your CAC, and you build a spreadsheet: total marketing spend divided by new customers, for the quarter. The number comes out somewhere between mildly embarrassing and completely uninterpretable, because at seed stage almost none of the inputs that make CAC meaningful actually exist yet. You don't have enough customers for the ratio to be stable, you can't cleanly separate paid acquisition from founder-led sales calls, and last month's number and this month's number can move sharply just because one deal happened to close on the 30th instead of the 1st.
At seed, CAC is a number you can't compute honestly.
The textbook version of CAC assumes a repeatable motion: a known channel, a known conversion rate, enough volume that noise averages out. A ten-customer company has none of that. One enterprise deal that took a founder's personal network to close will wreck the average as badly as one campaign that happened to catch a good week. Reporting the number anyway, with a straight face, teaches the board the wrong lesson twice: once when it looks great because of an outlier, and once when it looks terrible for the same reason.
The honest move is to say what CAC actually requires, and to say plainly that a ten- or twenty-customer company doesn't have it yet. That is not evasion. It is the more rigorous answer, and boards that have sat through a few funding cycles usually respect it more than a confident number that falls apart under one follow-up question.
Payback on paid spend is the number that holds up.
What does hold up at seed is a narrower question: for the dollars you spent on paid channels specifically, in a defined window, how long until a customer's revenue covers what it cost to get them. This isolates the one input you actually control and measure cleanly, ad spend, from the ones you can't yet separate, like founder time or inbound that arrived because someone mentioned you at a dinner.
Payback answers the question a board actually needs answered. If we put another dollar into this channel, roughly how long before it comes back. That is a decision-relevant number even with ten data points, because it is asking about a mechanism, not an average across your whole go-to-market motion.
Treat every input as a range, and recheck it every quarter.
A ten-customer company's payback period is not a fact. It is an estimate with wide error bars, and the discipline that matters is admitting that in the same sentence you state the number. Say the range, say what is driving the uncertainty, a lumpy sales cycle, a channel that has only run for six weeks, and commit to rechecking it next quarter rather than treating this quarter's number as settled.
This is also where the measurement question actually gets decided: not by which tool you buy, but by whether anyone is rereading the payback number often enough to notice when the range narrows and the estimate becomes something closer to a fact. A board report that tracks the same range quarter over quarter shows more discipline than one that presents a new precise figure each time.
The point is which lever to pull next, not a report that looks finished.
A board update that says "CAC is a fixed number" invites a discussion about whether the number is good or bad, which is usually the wrong discussion at this stage. A board update that says "paid payback is trending toward a range, here's what's driving the trend, here's why the range is still wide" invites a discussion about what to do next: spend more here, less there, wait for another quarter of data. That is the version worth having in the room.
The number a startup at seed can defend is narrower than the one the templates ask for, and it says more. It also ages better. A single CAC figure from a ten-customer quarter is stale the moment the eleventh customer closes; a payback range that gets rechecked every quarter is designed to be revised, so revising it never looks like a mistake.
Boards that have seen a few seed rounds recognize the difference immediately, and the founders who lead with the narrower, honest number tend to spend less of the meeting defending it and more of it discussing what to do next.
We build the reporting stack around the question a founder actually needs to answer next, not around a metric that looks complete before it has earned that. Tell us what you're building.